Lesson 2. Use tidyverse pipes to subset time series data in R

In this lesson, we will learn how to import a larger dataset, and test our skills cleaning and plotting the data.

Learning Objectives

After completing this tutorial, you will be able to:

  • Subset data using the dplyr filter() function
  • Use dplyr pipes to manipulate data in R
  • Describe what a pipe does and how it is used to manipulate data in R

What you need

You need R and RStudio to complete this tutorial. Also we recommend that you have an earth-analytics directory setup on your computer with a /data directory within it.

R Libraries to Install:

  • ggplot2: install.packages("ggplot2")
  • dplyr: install.packages("dplyr")
  • lubridate: install.packages("lubridate")

Download Week 2 Data

Important - Data Organization

Before you begin this lesson, be sure that you’ve downloaded the dataset above. You will need to UNZIP the zip file. When you do this, be sure that your directory looks like the image below: note that all of the data are within the week2 directory. They are not nested within another directory. You may have to copy and paste your files to make this look right.

week 2 file organization
Your `week_02` file directory should look like the one above. Note that the data directory is directly under the earth-analytics folder.

Get started with time series data

To begin, load the ggplot2 and dplyr libraries. Also, set your working directory. Finally, set stringsAsFactors to FALSE globally using options(stringsAsFactors = FALSE).

# set your working directory to the earth-analytics directory
# setwd("working-dir-path-here")

# load packages
## Attaching package: 'lubridate'
## The following object is masked from 'package:base':
##     date

# set strings as factors to false
options(stringsAsFactors = FALSE)

Import precipitation time series

We will use a precipitation dataset collected by the National Centers for Environmental Information (formerly National Climate Data Center) Cooperative Observer Network (COOP) station 050843 in Boulder, CO. The data cover the time span between 1 January 2003 through 31 December 2013.

To begin, use read.csv() to import the .csv file.

# download the data
# download.file(url = "https://ndownloader.figshare.com/files/7283285",
#              destfile = "data/week_02/805325-precip-dailysum_2003-2013.csv")

# import the data
boulder_daily_precip <- read.csv("data/week_02/precipitation/805325-precip-dailysum-2003-2013.csv",
         header = TRUE)

# view first 6 lines of the data
## 1 1/1/03         0.00 COOP:050843 BOULDER 2 CO US    1650.5 40.03389
## 2 1/5/03       999.99 COOP:050843 BOULDER 2 CO US    1650.5 40.03389
## 3 2/1/03         0.00 COOP:050843 BOULDER 2 CO US    1650.5 40.03389
## 4 2/2/03       999.99 COOP:050843 BOULDER 2 CO US    1650.5 40.03389
## 5 2/3/03         0.40 COOP:050843 BOULDER 2 CO US    1650.5 40.03389
## 6 2/5/03         0.20 COOP:050843 BOULDER 2 CO US    1650.5 40.03389
## 1 -105.2811 2003      1
## 2 -105.2811 2003      5
## 3 -105.2811 2003     32
## 4 -105.2811 2003     33
## 5 -105.2811 2003     34
## 6 -105.2811 2003     36

# view structure of data
## 'data.frame':	792 obs. of  9 variables:
##  $ DATE        : chr  "1/1/03" "1/5/03" "2/1/03" "2/2/03" ...
##  $ DAILY_PRECIP: num  0e+00 1e+03 0e+00 1e+03 4e-01 ...
##  $ STATION     : chr  "COOP:050843" "COOP:050843" "COOP:050843" "COOP:050843" ...
##  $ ELEVATION   : num  1650 1650 1650 1650 1650 ...
##  $ LATITUDE    : num  40 40 40 40 40 ...
##  $ LONGITUDE   : num  -105 -105 -105 -105 -105 ...
##  $ YEAR        : int  2003 2003 2003 2003 2003 2003 2003 2003 2003 2003 ...
##  $ JULIAN      : int  1 5 32 33 34 36 37 38 41 49 ...

# are there any unusual / No data values?
##     Min.  1st Qu.   Median     Mean  3rd Qu.     Max. 
##    0.000    0.100    0.100    5.297    0.300  999.990
## [1] 999.99

About the Data

Viewing the structure of these data, we can see that different types of data are included in this file:

  • STATION and STATION_NAME: Identification of the COOP station.
  • ELEVATION, LATITUDE and LONGITUDE: The spatial location of the station.
  • DATE: The date when the data were collected in the format: YYYYMMDD. Notice that DATE is currently class chr, meaning the data is interpreted as a character class and not as a date.
  • DAILY_PRECIP: The total precipitation in inches. Important: the metadata notes that the value 999.99 indicates missing data. Also important, hours with no precipitation are not recorded.
  • YEAR: The year the data were collected.
  • JULIAN: The JULIAN DAY the data were collected.

Additional information about the data, known as metadata, is available in the PRECIP_HLY_documentation.pdf. The metadata tell us that the noData value for these data is 999.99. IMPORTANT: We have modified these data a bit for ease of teaching and learning. Specifically, we’ve aggregated the data to represent daily sum values and added some noData values to ensure you learn how to clean them!

You can download the original complete data subset with additional documentation here.

Optional challenge

Using everything you’ve learned in the previous lessons:

  • Import the dataset: data/week_02/precipitation/805325-precip-dailysum-2003-2013.csv
  • Clean the data by assigning noData values to NA
  • Make sure the date column is a date class
  • When you are done, plot it using ggplot()
    • Be sure to include a TITLE, and label the X and Y axes
    • Change the color of the plotted points

Some notes to help you along:

  • Date: Be sure to take off the date format when you import the data.
  • NoData Values: We know that the no data value = 999.99. We can account for this when we read in the data. Remember how?

Your final plot should look something like the plot below.

Import data and reassign na values

To begin, import the data. Be sure to use the na.strings argument to remove NA values. Also our data have a header (the first row represents column names) so set header = TRUE

# import data
boulder_daily_precip <- read.csv("data/week_02/precipitation/805325-precip-dailysum-2003-2013.csv",
         header = TRUE,
         na.strings = 999.99)

Next, take care of the date field. In this case we have month/day/year. We can use ?strptime to figure out which letters we need to use in the format = argument to ensure our data elements (month, day and year) are understood by R.

In this case we want to use

  • %m - for month
  • %d - for day
  • %y - for year

Also take note of the format of our date. In this case, each date element is separated by a /.

# format date field
boulder_daily_precip$DATE <- as.Date(boulder_daily_precip$DATE,
                                     format = "%m/%d/%y")

Finally, we can plot the data using ggplot(). Notice that when we plot, we first populate the data and aes (aesthetics).

  • data = contain the data frame that we want to plot
  • aes = contain the x and y variables that we want to plot.

Finally geom_point() represents the geometry that you want to plot. In this case you are creating a scatter plot (using points).

# plot the data using ggplot2
ggplot(data=boulder_daily_precip, aes(x = DATE, y = DAILY_PRECIP)) +
      geom_point() +
      labs(title = "Precipitation - Boulder, Colorado")

precip plot w fixed dates

NA values and warnings

When we plot the data, we get a warning that says:

## Warning: Removed 4 rows containing missing values (geom_point).

We can get rid of this warning by removing NA or missing data values from our data. A warning is just R’s way of letting you know that something may be wrong. In this case, it can’t plot 4 data points because there are missing data values there.

Let’s remove the missing data value rows using a dplyr pipe and the na.omit() function. We will talk about pipes in just a minute!

boulder_daily_precip <- boulder_daily_precip %>%

Now we can plot the data without a warning!

# plot the data using ggplot2
ggplot(data=boulder_daily_precip, aes(x = DATE, y = DAILY_PRECIP)) +
      geom_point(color = "darkorchid4") +
      labs(title = "Precipitation - Boulder, Colorado")

precip plot w fixed dates and no na values

Don’t forget to add x and y axis labels to your plot! Use the labs() function to add a title, x and y label (and subtitle if you’d like) to your plot.

labs(title = "Hourly Precipitation - Boulder Station\n 2003-2013",
     x = "Date",
     y = "Precipitation (Inches)")
ggplot(data = boulder_daily_precip, aes(DATE, DAILY_PRECIP)) +
      geom_point(color = "darkorchid4") +
      labs(title = "Hourly Precipitation - Boulder Station\n 2003-2013",
           x = "Date",
           y = "Precipitation (Inches)")

plot with titles and labels

You can add a ggplot theme to adjust the look of your plot quickly too. Below we use theme_bw(). Below we also adjust the base font size to make the labels a bit larger base_size = 11.

Data Tip: Learn more about built in ggplot themes

ggplot(data = boulder_daily_precip, aes(DATE, DAILY_PRECIP)) +
      geom_point(color = "darkorchid4") +
      labs(title = "Hourly Precipitation - Boulder Station\n 2003-2013",
           x = "Date",
           y = "Precipitation (Inches)") + theme_bw(base_size = 11)

plot with titles and labels black and white

Data Tip: For a more thorough review of date/time classes, see the NEON tutorial Dealing With Dates & Times in R - as.Date, POSIXct, POSIXlt.

Optional challenge

Take a close look at the plot:

  • What does each point represent?
  • Use the min() and max() functions to determine the minimum and maximum precipitation values for the 10 year span?

Introduction to the pipe %<%

Above we used pipes to manipulate our data. Specifically we removed NA values in a pipe with na.omit().

Pipes let you take the output of one function and send it directly to the next, which is useful when you need to do many things to the same data set. Pipes in R look like %>% and are made available via the magrittr package, installed automatically with dplyr.

boulder_daily_precip <- boulder_daily_precip %>%

Pipes are nice to use when coding because:

  1. they remove intermediately created variables (keeping our environment cleaner / fewer variables are saved memory)
  2. they combine multiple steps of processing into a clean set of steps that is easy to read once you become familiar with the pipes syntax

We can do all of the same things that we did above with one pipe. Let’s see how:

# format date field without pipes
boulder_daily_precip$DATE <- as.Date(boulder_daily_precip$DATE,
                                     format = "%m/%d/%y")

With pipes we can use the mutate function to either create a new column or modify the format or contents of an existing column.

boulder_daily_precip <- boulder_daily_precip %>%
  mutate(DATE = as.Date(DATE, format = "%m/%d/%y"))

We can then add the na.omit() function to the above code

boulder_daily_precip <- boulder_daily_precip %>%
  mutate(DATE = as.Date(DATE, format = "%m/%d/%y")) %>%

Notice that each time you assign the pipe to a variable, you are overwriting that variable.

boulder_daily_precip <- boulder_daily_precip

In this case we are just updating our current boulder_daily_precip variable.

The process above avoids processing the data in separate steps, and potentially creating new variables each time. We can even send the output to ggplot(). When we send output to ggplot() in a pipe, we don’t need the use the data argument (data = boulder_daily_precip) because we send the data throught the pipe. Like this:

Note: that because you are creating a plot with the code below, you don’t need to assign the pipe to a variable. Thus you leave out the

boulder_daily_precip <-


boulder_daily_precip %>%
  mutate(DATE = as.Date(boulder_daily_precip$DATE, format = "%m/%d/%y")) %>%
  na.omit() %>%
ggplot(aes(DATE, DAILY_PRECIP)) +
      geom_point(color = "darkorchid4") +
      labs(title = "Hourly Precipitation - Boulder Station\n 2003-2013",
           subtitle = "plotted using pipes",
           x = "Date",
           y = "Precipitation (Inches)") + theme_bw()

data plotted with a pipe

Subset the data

You may want to only work with a subset of your time series data. Let’s create a subset of data for the time period around the flood between 15 August to 15 October 2013. We use the filter() function in the dplyr package to do this and pipes!

# subset 2 months around flood
precip_boulder_AugOct <- boulder_daily_precip %>%
                        filter(DATE >= as.Date('2013-08-15') & DATE <= as.Date('2013-10-15'))

In the code above, we use the pipe to send the boulder_daily_precip data through a filter step. In that filter step, we filter out only the rows within the date range that we specified. Since %>% takes the object on its left and passes it as the first argument to the function on its right, we don’t need to explicitly include it as an argument to the filter() function.

# check the first & last dates
## [1] "2013-08-21"
## [1] "2013-10-11"

# create new plot
ggplot(data = precip_boulder_AugOct, aes(DATE,DAILY_PRECIP)) +
  geom_bar(stat = "identity", fill = "darkorchid4") +
  xlab("Date") + ylab("Precipitation (inches)") +
  ggtitle("Daily Total Precipitation Aug - Oct 2013 for Boulder Creek") +

precip plot subset

Optional challenge

Create a subset from the same dates in 2012 to compare to the 2013 plot. Use the ylim() argument to ensure the y axis range is the SAME as the previous plot - from 0 to 10”.

How different was the rainfall in 2012?

HINT: type ?lims in the console to see how the xlim and ylim arguments work.

precip plot subset 2

Leave a Comment